Your bestseller might be your least profitable SKU

Every brand knows what its bestseller sells. Almost nobody knows what it earns. Revenue rank sits on the front page of every dashboard, updated hourly. Contribution margin (gross margin minus the ad spend, return cost, and fulfillment cost attributable to that product) usually sits nowhere at all.

The pattern I see over and over with founders at this size is that they can quote their hero SKU's revenue to the dollar, off the top of their head, with no notes. Ask what it contributes after ads and returns and freight, and the room goes quiet. That gap is the whole article.

Revenue rank is not profit rank

Let me be straight about what this piece does and does not claim, because the internet is full of confident headlines here.

There is no published, cross-brand dataset proving that a typical brand's single highest-revenue SKU lands in the bottom quartile of its catalog by contribution margin. Nobody has measured that reversal at that level of precision. If someone quotes you a percentage of brands where the bestseller is the worst SKU, ask them for the study. There isn't one.

What is documented, separately and by name, is every cost lever that would produce that reversal. Paid ad spend concentrates on products that are already winning, because that is what the algorithms are built to do. Return rates are highest in apparel and footwear, exactly the categories where hero-SKU branding is most common. Bracketing gets worse on popular, fast-moving products specifically. And dimensional-weight freight rules penalize bulky-but-light items with total indifference to how well they sell.

Stack those four together and you have a mechanism. A mechanism is not proof that your bestseller is underwater. It is a reason to check, and checking takes an afternoon.

The load-bearing fact of this article is that a large share of every catalog is quietly unprofitable already. The prior probability that one of your SKUs is a margin loser is not low. It is high. Revenue rank simply gives you no information about whether that SKU is one of them.

The three places a hero SKU's margin leaks

Ad spend goes where the clicks are, not where the margin is. Catalog and dynamic-ad algorithms optimize toward conversion probability, not contribution margin, because contribution margin is a number you never gave them. They cannot optimize for what they cannot see. Meta's share of tracked ecommerce ad spend reached 68.3% in 2025, per Triple Whale's 2025 Ecommerce Benchmarks Report, built on $18.4B in tracked spend across more than 33,000 brands. That is a lot of budget flowing through systems ranking your products by click likelihood. Ecommerce.co.za reported in 2025 that 40% of Google Shopping ad spend generates zero revenue, on the same logic, though that figure comes from one agency's own client work rather than an independent study. Treat that one as directional, but the mechanism is not controversial: your hero SKU wins the auction for your own budget because it converts, and converting is not earning.

Returns are highest exactly where hero SKUs live. NRF and Happy Returns put the 2025 US online return rate at 19.3%, against 15.8% for retail overall. Appriss Retail and Deloitte measured online returns at 24.52% of online sales in 2024. Apparel runs higher still, roughly 20% to 40% depending on whose synthesis you read, with footwear around 18% to 30% and general merchandise closer to 14% to 18%.

The strongest confirmation is not a vendor benchmark. It is a public filing. FIGS, Inc., which sells medical scrubs, told the SEC in its fiscal-2025 10-K that "our product portfolio has resulted in a return rate of approximately 10% from 2021 through 2025, which is far lower than the broader online apparel return rates that tend to be in the 30% to 40% range." A public apparel company disclosed its return rate as a competitive advantage, even though it operates in one of the categories least exposed to bracketing, since scrubs are functional workwear bought by repeat customers who already know their size. That only makes sense if the category default is a serious drag.

Volume makes this worse. Narvar found 63% of shoppers bracket their orders, buying multiple sizes or colors intending to send most back, up from 55% in 2019, and that size and fit drive 45% of all returns. Optoro's 2024 report calls bracketing especially problematic for popular, fast-moving SKUs, because it erodes the reliable available-to-sell inventory counts retailers depend on, and it found shoppers are most likely to embrace bracketing when shopping for apparel and accessories (35%), the exact category most hero SKUs live in. We go deeper on what returns actually cost, once restocking, write-offs, and lost inventory are counted, in the true cost of apparel returns.

Freight does not care what your bestseller is. Carriers bill on dimensional weight when the box is bigger than the contents are heavy. UPS and FedEx both use a divisor of 139 on contract accounts, and as of August 18 2025 both round every fractional inch up before running the calculation, so a box measuring 11.1 inches on a side gets billed as 12. Jay Group's parcel-consulting analysis works the example: a carton measured at 11.1 x 8.5 x 6.2 inches now rounds to 12 x 9 x 7, moving it from roughly 5 pounds of billed weight to roughly 6, about 20% more, with zero physical change to the package.

Carrier	DIM divisor	Scope	Rounding rule
UPS (contract/daily)	139	All domestic parcels	Rounds each fractional inch up (from Aug 18 2025)
UPS (retail/counter)	166	Retail and counter shipments	Same rounding rule applies
FedEx (Ground/Express)	139	All domestic parcels	Rounds each fractional inch up (from Aug 18 2025)
USPS (before July 12 2026)	166	Parcels over 1 cubic foot only	Standard rounding
USPS (current, since July 12 2026)	139	Parcels over 1 cubic foot	Ceiling rounding adopted

When we build this out with operators, the line that surprises people is almost never the return rate. Everyone knows returns hurt. It is the freight line, because it moved without anyone sending an email about it, and because a packaging decision made two years ago is still quietly setting the bill every day.

Run the math: an illustrative contribution-margin bridge

Here is the arithmetic. This is a model, not observed data from any brand. Every input is published and cited, and every figure reproduces exactly from the stated assumptions. Take a hypothetical apparel hero SKU doing $1,000,000 a year in revenue, and swap your own numbers in.

Line	$ on $1M SKU revenue	Where the input comes from
Gross margin at 57%	$570,000	NYU Stern apparel industry average (Damodaran, January 2026 data). Benchmark.
Less: allocated ad spend at 15% of revenue	-$150,000	Stated modeling assumption for a hero SKU defending its rank. Not a benchmark.
Less: return processing (24.5% return rate x 26.5% of item value)	-$64,925	Appriss Retail 2024 (24.52% online, all categories) x Optoro 2024 ($26.50 per $100). Both benchmarks. Uses the all-category rate, not the higher apparel-specific range, so this input runs conservative.
Less: outbound fulfillment and freight at 15% of revenue	-$150,000	Stated modeling assumption. Not a benchmark. Deducted in full, because industry gross margin is measured before outbound shipping (see the note below the bridge).
Less: DIM rounding premium (15% increase on that freight)	-$22,500	The 15% increase is the midpoint of a reported 8-22% range. Directional.
Contribution margin remaining	$182,575 (18.3%)	570,000 - 150,000 - 64,925 - 150,000 - 22,500

One line in that bridge deserves a note, because it is where most napkin versions of this math go wrong. The freight baseline gets deducted in full, not just the DIM premium on top of it. That is deliberate. Industry gross margin is measured before outbound shipping: FIGS, to use a filer already cited here, tells the SEC that cost of goods sold "consists principally of the cost of purchased merchandise and includes import duties, tariffs and other taxes, freight-in," while "shipping and handling costs are associated with outbound freight after control over a product has transferred to a customer and, as such, are included in selling expenses." Freight-in sits above the gross-margin line. Freight-out sits below it. So a 57% gross margin has not paid to get the box to anybody, and a bridge that deducts only the DIM delta is quietly counting the other $150,000 as free.

Notice what this bridge does not do. It does not land the SKU underwater. It lands at 18.3%: thin, but positive, and survivable for plenty of brands. That is the honest result of the honest model, and it matters more than a scarier number would.

The point is the distance travelled. A 57% gross margin became 18.3% before a dollar of overhead, salary, software, or rent was paid. Nearly 39 points left the building through four lines most brands never allocate to the SKU at all. Now change one input. Push the ad allocation to 25% to defend rank and it drops to 8.3%. Take the return rate to the top of the apparel range instead of the all-category middle and it lands at 4.2%. None of those are exotic assumptions. They are Tuesday.

A 57% gross margin turning into an 18.3% contribution margin is not a scandal. It is arithmetic. The scandal is that most brands never run it, so they cannot tell the difference between a hero SKU at 18% and one at 3%, and they make the same decision about both.

A big share of every catalog earns nothing

This is where two more data points point the same direction, even though neither one is airtight on its own, and it is why "go check" is not a hedge.

Linnworks states, without citing an underlying study, that 62% of ecommerce SKUs are not profitable once fully loaded costs are counted, naming advertising, returns, and fulfillment as the three drains teams miss when they focus on top-line performance. PwC's Strategy& shows the same pattern in a 2019 illustrative example from one global CPG company's product portfolio: half its SKUs drove less than 5% of gross margin, and 35% drove zero incremental profitability.

Neither is a survey of many companies. One is a vendor claim with no cited methodology; the other is an illustrative example from a single CPG portfolio, not an ecommerce-specific sample. But they point the same direction: something like a third to two-thirds of a catalog earning nothing is not an outlier reading. (A third figure circulates here: 32% of SKUs with active ad spend generate zero revenue, which Linnworks attributes to Conjura, an ecommerce analytics platform. We read it on Linnworks rather than in a Conjura publication, and it is one vendor's platform data rather than an independent study, so treat it as a data point, not a benchmark.)

Here is the connection people miss. Neither source says the unprofitable SKUs are the slow movers. That is an assumption everyone imports for free, because it feels right. Revenue concentration in DTC catalogs is steep. In the same PwC/Strategy& portfolio cited above, Class A SKUs, 28% of the total, drove 80% of cumulative gross margin, a shape that shows up consistently in case-level Pareto analyses of ecommerce catalogs. So if a chunk of your catalog is unprofitable and you have never checked which chunk, the SKU with the most units flowing through it is where being wrong costs the most.

What to do about it

Get ad spend to the SKU level, even imperfectly. Blended ROAS is the single biggest reason nobody knows their hero SKU's real margin. If your catalog runs as one dynamic campaign, break your top few products out into their own campaigns for a month. You lose some algorithmic efficiency and gain the ability to answer the question. Imperfect and honest beats blended and useless.

Measure the box before you blame the product. Pull last month's parcel invoices and compare billed weight to actual weight on your top SKU. If billed is higher, you are paying a DIM penalty, and the fix is cardboard, not strategy. It is the cheapest margin available to most brands and needs no customer to change behavior.

Attack the return rate at the product page, not the returns policy. Size and fit drive 45% of returns (Narvar, 2022). Better size guidance, honest fit photography, and reviews that mention height and weight move that number. Tightening the returns policy moves conversion instead, usually the wrong way.

Look at price before you look at spend. When we've worked through this with founders, the move that works is almost never killing the SKU. A hero product does brand and discovery work the spreadsheet cannot see, and cutting its budget can take the rest of the catalog's traffic with it. A modest price increase on a product people already choose usually costs less volume than the model predicts, and it drops straight to contribution margin. For the fuller version of this math, our contribution margin guide for DTC brands walks the stack.

What this article is not claiming

We are not claiming your bestseller is your worst SKU, or that any particular share of brands has that problem. No one has published research measuring it, and we are not going to invent a number to make a headline work.

What the evidence establishes is narrower and more useful. Every cost lever that would flip a hero SKU from winner to margin loser is real, named, dated, and measurable on your own data. A large share of the average catalog is unprofitable once fully loaded, per two sources that point the same direction even though neither is a rigorous population study. Those costs concentrate on high-volume products. And almost nobody tracks contribution margin at the SKU level, so the reversal could sit in your catalog for years without a dashboard flinching.

That is enough to justify one afternoon with your own numbers. It is not enough to justify a stat you can quote at a conference, and you should be suspicious of anyone offering you one.

Related reading. For the same SKU-level margin math applied elsewhere, see return-rate profit math by SKU and the bundle margin check. For how we help brands model margin and cash, see our fractional CFO work.

Sources and methodology

Return-rate figures come from named annual returns measurements, not from us. The 2025 US online return rate of 19.3% (against 15.8% for retail overall) comes from NRF and Happy Returns' 2025 Retail Returns Landscape, published October 2025. The 24.52% online figure comes from Appriss Retail and Deloitte's 2024 Consumer Returns in the Retail Industry, published December 30 2024. Neither publishes a clean category-by-category table, so the bands here (apparel roughly 20-40%, footwear 18-30%, general merchandise 14-18%) are cross-source syntheses presented as ranges. The chart plots the median of each cross-source range, not a strict midpoint, so read the bars as ranges rather than precise measurements.

The strongest return-rate citation here is a public filing, not a benchmark. FIGS, Inc.'s fiscal-2025 Form 10-K (filed February 26 2026, CIK 0001846576) states verbatim that "our product portfolio has resulted in a return rate of approximately 10% from 2021 through 2025, which is far lower than the broader online apparel return rates that tend to be in the 30% to 40% range." FIGS's own ~10% figure is a hard, company-disclosed operating metric with SEC liability attached, which is why it leads the returns section and beats any vendor report. The 30-40% "broader online apparel" figure in the same sentence is a different kind of claim: FIGS is asserting it about other companies, unsourced, not measuring it, so treat it as directional corroboration of the benchmark syntheses above rather than independent proof of them. Other filers reviewed (A.K.A. Brands, Duluth Holdings, Dillard's) discuss returns only qualitatively and were not used as numeric sources.

SKU-profitability findings come from two sources of different weight. The 62% figure comes from a Linnworks blog post published April 25 2025, which states the figure without citing an underlying study or methodology; we use it as a named claim, not a verified research finding. The 35% and half-of-SKUs figures come from PwC/Strategy&'s "Unlocking Hidden Value in Product Portfolios" (2019), an illustrative example from one global CPG company's portfolio, not a survey of many companies or an ecommerce-specific sample. Shown together because they point the same direction, not because either is a rigorous population study. Neither measures whether a brand's highest-revenue SKU is among the unprofitable ones, which is precisely the gap this article refuses to paper over.

DIM-weight mechanics are confirmed; the cost-increase range is not. The 139 divisor, the USPS move from 166 to 139 in July 2026, and the August 18 2025 rounding change are confirmed against published carrier documentation and corroborated across Jay Group's parcel-consulting analysis and Supply Chain Dive's reporting. The 8-22% per-package cost-increase range is weaker: it traces to E-Commerce Times reporting via a secondhand citation and the methodology could not be independently confirmed. It is used only as the midpoint input to an explicitly illustrative model and flagged as directional wherever it appears.

The contribution-margin bridge is an illustrative model, and two of its five inputs are assumptions. Gross margin at 57% (NYU Stern apparel industry average, 56.88% as of January 2026) and return processing at 24.5% x 26.5% (Appriss Retail x Optoro) are published benchmarks. The 15% ad allocation and 15%-of-revenue freight baseline are stated modeling assumptions chosen to be plausible for a hero SKU, not measured values. The arithmetic reproduces exactly: 570,000 - 150,000 - 64,925 - 150,000 - 22,500 = 182,575 on $1,000,000 of revenue, or 18.3%. The full freight baseline is deducted, not just the DIM premium, because industry gross-margin datasets are measured before outbound shipping: outbound freight sits in selling expense rather than cost of goods sold for DTC apparel filers, as FIGS' 10-K states explicitly ("shipping and handling costs are associated with outbound freight after control over a product has transferred to a customer and, as such, are included in selling expenses"), while freight-in is inside COGS and therefore already reflected in the 57%. The two sensitivities quoted in that section also reproduce: at a 25% ad allocation, 570,000 - 250,000 - 64,925 - 150,000 - 22,500 = 82,575, or 8.3%; taking the return rate to 40% on top of that, 570,000 - 250,000 - 106,000 - 150,000 - 22,500 = 41,500, or 4.2%. No proprietary, client, or panel data is used anywhere in this article.

Bracketing behavior comes from Narvar's named 2022 survey. The 63% (up from 55% in 2019) and 45% size-and-fit figures come from Narvar's "State of Returns: The End of One-Size-Fits-All Returns", a survey of 2,023 consumers who had returned an online purchase in the prior six months. It is a 2022 survey used here as the best available named source on bracketing prevalence, not a current-year figure. Optoro's 2024 State of Returns report finds a similar order of magnitude two years fresher (63% of shoppers reporting they purposely over-purchase intending to return some items, up from 58% the prior year), though the two surveys use different baseline years and are not merged into a single figure here.

Ad-spend concentration is a documented mechanism, not a measured distribution. Meta's 68.3% share of tracked ecommerce ad spend comes from Triple Whale's 2025 Ecommerce Benchmarks Report, built on $18.4B in tracked spend across more than 33,000 brands. The 40%-of-Google-Shopping-spend figure comes from an agency comment published on Ecommerce.co.za (2025), drawn from that agency's own client work rather than an independent study, and is directional. No platform-authored dataset quantifying ad spend by SKU rank exists publicly, so this article describes the mechanism and declines to attach a prevalence number to it.

Frequently asked questions

how do i actually calculate contribution margin for one sku?

Start with that SKU's revenue, subtract its cost of goods, then subtract the ad spend you can attribute to it, the cost of processing its returns, and its actual fulfillment and freight cost. What's left is contribution margin. The hard part isn't the formula, it's getting ad spend and freight down to the SKU level instead of blended across the catalog.

is my bestseller actually losing me money?

Nobody can tell you that from published data, including us. What published data does tell you is that Linnworks puts the share of ecommerce SKUs that are unprofitable once fully loaded at 62% (2025, without citing an underlying study), and that the cost lines which hit hardest all concentrate on high-volume products. It's a one-afternoon calculation on your own numbers. Run it before you assume either way.

why does my top selling product have the highest return rate?

Partly volume and partly bracketing. Narvar found 63% of shoppers buy multiple sizes or colors intending to return most of them, and size and fit drive 45% of all returns. Optoro's 2024 report calls bracketing especially problematic for popular, fast-moving SKUs because it erodes reliable available-to-sell inventory counts, and found shoppers are most likely to bracket when shopping for apparel and accessories (35%), the exact category most hero SKUs live in.

what is dimensional weight shipping and why did my shipping bill go up?

Carriers bill you on whichever is greater: actual weight, or length times width times height divided by a divisor. UPS and FedEx use 139 for contract accounts. Since August 18 2025 both round every fractional inch up before doing that math, so a box measuring 11.1 inches gets billed as 12. Nothing about your product changed. The billing rule did.

should i stop advertising my bestseller if the margin is bad?

Usually no, and that's the trap. Cutting spend on a hero SKU can take the rest of the catalog's discovery with it. Look at price, packaging dimensions, and return rate first, because those fix the margin without touching the top of the funnel. Pulling spend is the last lever, not the first.

how much does it really cost to process a return?

Optoro's 2024 measurement is $26.50 per $100 of returned merchandise, or about 26.5% of item value, and that excludes what you lose when the item comes back unsellable or has to be marked down. Treat 26.5% as the floor, not the ceiling.

what percentage of my skus are probably unprofitable?

Two different sources put it high, though neither is a survey of many companies. Linnworks states, without citing an underlying study, that 62% of ecommerce SKUs are not profitable once ad spend, returns and fulfillment are counted. PwC's Strategy& showed 35% of SKUs driving zero incremental profitability and half driving under 5% of gross margin, in a 2019 illustrative example from one global CPG company. Neither tells you which of yours, which is the whole point of running it.